{"payload":{"feedbackUrl":"https://github.com/orgs/community/discussions/53140","repo":{"id":690828010,"defaultBranch":"default","name":"stablehlo","ownerLogin":"iree-org","currentUserCanPush":false,"isFork":true,"isEmpty":false,"createdAt":"2023-09-13T00:58:44.000Z","ownerAvatar":"https://avatars.githubusercontent.com/u/107954215?v=4","public":true,"private":false,"isOrgOwned":true},"refInfo":{"name":"","listCacheKey":"v0:1701990592.0","currentOid":""},"activityList":{"items":[{"before":"dd48ec58d3bb8d674adf56715d4394102538fa84","after":"57d16b101fb8b71bee13fa93e320d0cccef87f52","ref":"refs/heads/main","pushedAt":"2024-06-21T15:45:22.000Z","pushType":"push","commitsCount":12,"pusher":{"login":"nirvedhmeshram","name":"Nirvedh Meshram","path":"/nirvedhmeshram","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/96096277?s=80&v=4"},"commit":{"message":"Add testdata files for quantized stablohlo programs (#2404)\n\nAdd quantized tests for the following stablehlo operations\r\n```\r\nabs\r\ncbrt\r\nceil\r\ncosine\r\ndivide\r\nexponential\r\nexponential_minus_one\r\nfloor\r\nlog\r\nlog_plus_one\r\nlogistic\r\nmultiply\r\nnegate\r\nsign\r\nsine\r\nsqrt\r\nsubtract\r\ntanh\r\n```\r\n\r\n## A few details\r\n- These tests are all auto-generated by a `testdata-generator` tool\r\nwhich takes an input a non-quantized program file from\r\nhttps://github.com/openxla/stablehlo/tree/main/stablehlo/tests/interpret\r\nor https://github.com/openxla/stablehlo/tree/main/stablehlo/testdata and\r\ngenerates a quantized program file in the same format as the input file.\r\n- The test file names have 1-1- correspondence with its input file name\r\nappended with `_qi8` appended, suggesting that all the quantized program\r\ninvolves `i8` as storage type.\r\n- The constant argument fed to each test file is the same as that fed to\r\nthe input program.\r\n\r\n \r\n## A few limitations\r\n1. Currently we are using the [Stablehlo\r\nQuantizer](https://github.com/tensorflow/tensorflow/tree/master/tensorflow/compiler/mlir/quantization/stablehlo)\r\nto generate quantized program which currently supports quantization of\r\n`f32` floating point programs only. In future this case be improved.\r\n- **Corollary**: If the input test files have non-f32 based pograms,\r\nthen it will be preserved unquantized in the output file. That is why we\r\ncan see some of the generated files have non-quantized programs as well.\r\n3. While scraping the input programs from\r\nhttps://github.com/openxla/stablehlo/tree/main/stablehlo/tests/interpret\r\nor https://github.com/openxla/stablehlo/tree/main/stablehlo/testdata we\r\nchoose only those programs which tests single stablehlo operations. This\r\nwhat the `testdata-generator` tool currently supports, but in very near\r\nfuture we can expect the tool to work on input test files with mutiple\r\noperations being tested.\r\n\r\n\r\nKeep an eye on more quantized tests to be populated!","shortMessageHtmlLink":"Add testdata files for quantized stablohlo programs (openxla#2404)"}},{"before":"3dce5fdfef144a8d9eff63a2f8adf0f4e5e045d6","after":"dd48ec58d3bb8d674adf56715d4394102538fa84","ref":"refs/heads/main","pushedAt":"2024-06-13T01:46:18.000Z","pushType":"push","commitsCount":32,"pusher":{"login":"antiagainst","name":"Lei Zhang","path":"/antiagainst","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/487928?s=80&v=4"},"commit":{"message":"Integrate LLVM at llvm/llvm-project@c012e487b724 (#2392)","shortMessageHtmlLink":"Integrate LLVM at llvm/llvm-project@c012e487b724 (openxla#2392)"}},{"before":"fe1b3a190411ead0821b1333662d96c802013f0d","after":"3dce5fdfef144a8d9eff63a2f8adf0f4e5e045d6","ref":"refs/heads/main","pushedAt":"2024-05-16T18:06:19.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"github-actions[bot]","name":null,"path":"/apps/github-actions","primaryAvatarUrl":"https://avatars.githubusercontent.com/in/15368?s=80&v=4"},"commit":{"message":"Regenerated pass documentation to fix bazel build failure at head (#2352)\n\ndocumentation is outdated after\r\nhttps://github.com/openxla/stablehlo/pull/2253\r\n\r\nValidated that the change actually fixes the failure\r\n\r\n\r\nFollow up: why https://github.com/openxla/stablehlo/pull/2253 CI check\r\ndid not complain?","shortMessageHtmlLink":"Regenerated pass documentation to fix bazel build failure at head (op…"}},{"before":"488749da83d1b3008a1750292f6d4d7449975261","after":"fe1b3a190411ead0821b1333662d96c802013f0d","ref":"refs/heads/main","pushedAt":"2024-05-16T15:05:20.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"github-actions[bot]","name":null,"path":"/apps/github-actions","primaryAvatarUrl":"https://avatars.githubusercontent.com/in/15368?s=80&v=4"},"commit":{"message":"Small nit fix in gather/scatter batching index computation in evaluator (#2351)\n\nUse `startIndicesIndex` in Ops.cpp for computing gather/scatter batching\r\nindex instead of `batchIndex`","shortMessageHtmlLink":"Small nit fix in gather/scatter batching index computation in evaluat…"}},{"before":"b217c4628be6b5f336e889458323fa24a017201f","after":"488749da83d1b3008a1750292f6d4d7449975261","ref":"refs/heads/main","pushedAt":"2024-05-16T03:05:37.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"github-actions[bot]","name":null,"path":"/apps/github-actions","primaryAvatarUrl":"https://avatars.githubusercontent.com/in/15368?s=80&v=4"},"commit":{"message":"Extract folders from StablehloRefineShapes into new StablehloAggressiveFolder pass (#2253)\n\nExtract folders from StablehloRefineShapes into new\r\nStablehloAggressiveFolder pass.\r\n\r\nThis patch addresses the problem of adding new folders and allowing\r\npotentially lossy computations in folding patterns. It preserves the\r\nbehavior of existing models and workflows that rely on shape refinement\r\nto fold shapes while assuming non-shape-related computations remain\r\nunchanged.","shortMessageHtmlLink":"Extract folders from StablehloRefineShapes into new StablehloAggressi…"}},{"before":"2d294ae07bab5ad197e393461d4038fef404a037","after":"b217c4628be6b5f336e889458323fa24a017201f","ref":"refs/heads/main","pushedAt":"2024-05-15T21:05:45.000Z","pushType":"push","commitsCount":2,"pusher":{"login":"github-actions[bot]","name":null,"path":"/apps/github-actions","primaryAvatarUrl":"https://avatars.githubusercontent.com/in/15368?s=80&v=4"},"commit":{"message":"Use snake_case instead of PascalCase for generated docs (#2349)\n\nIn an effort to be consistent with the rest of the documentation.","shortMessageHtmlLink":"Use snake_case instead of PascalCase for generated docs (openxla#2349)"}},{"before":"6051bcdfc057bba9a572da01b413106b4bfa3161","after":"2d294ae07bab5ad197e393461d4038fef404a037","ref":"refs/heads/main","pushedAt":"2024-05-15T19:28:32.000Z","pushType":"push","commitsCount":6,"pusher":{"login":"stellaraccident","name":"Stella Laurenzo","path":"/stellaraccident","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/494208?s=80&v=4"},"commit":{"message":"Make clang-format errors more helpful (#2330)\n\nSpecifically:\r\n- Provide a command that can be used to fix formatting problems\r\nautomatically\r\n- Make it obvious what version of clang-format to use\r\n\r\nAlso:\r\n- Don't specify --style=google because we have a .clang-format file that\r\nalready specifies that.\r\n- Only show the version if there are formatting failures\r\n\r\nCloses https://github.com/openxla/stablehlo/issues/2223.","shortMessageHtmlLink":"Make clang-format errors more helpful (openxla#2330)"}},{"before":"94b6ffbba4bb96ef2073f19369e3bc175f1484a9","after":"6051bcdfc057bba9a572da01b413106b4bfa3161","ref":"refs/heads/main","pushedAt":"2024-05-14T21:24:53.000Z","pushType":"push","commitsCount":21,"pusher":{"login":"stellaraccident","name":"Stella Laurenzo","path":"/stellaraccident","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/494208?s=80&v=4"},"commit":{"message":"Update constraint comment for consistency (#2343)","shortMessageHtmlLink":"Update constraint comment for consistency (openxla#2343)"}},{"before":"1431e1c26eb2a6b2fd8cdcb4f7dc5ce6936cf527","after":"94b6ffbba4bb96ef2073f19369e3bc175f1484a9","ref":"refs/heads/main","pushedAt":"2024-05-10T18:10:15.000Z","pushType":"push","commitsCount":2,"pusher":{"login":"github-actions[bot]","name":null,"path":"/apps/github-actions","primaryAvatarUrl":"https://avatars.githubusercontent.com/in/15368?s=80&v=4"},"commit":{"message":"Derive adaptor from OpTy in linalg dot lowering template (#2322)\n\nFollow-up to\r\nhttps://github.com/openxla/stablehlo/pull/2319#discussion_r1597007257","shortMessageHtmlLink":"Derive adaptor from OpTy in linalg dot lowering template (openxla#2322)"}},{"before":"49d93822e1301dc8f29d88f1af0c989ed8da4ee3","after":"1431e1c26eb2a6b2fd8cdcb4f7dc5ce6936cf527","ref":"refs/heads/main","pushedAt":"2024-05-10T17:59:01.000Z","pushType":"push","commitsCount":12,"pusher":{"login":"github-actions[bot]","name":null,"path":"/apps/github-actions","primaryAvatarUrl":"https://avatars.githubusercontent.com/in/15368?s=80&v=4"},"commit":{"message":"Add spec, interpreter, and verifier for DynamicConvOp (#2312)","shortMessageHtmlLink":"Add spec, interpreter, and verifier for DynamicConvOp (openxla#2312)"}},{"before":"b2e80ee330933ad9e92c24cab54eb63c2a5c18b9","after":"1c1f089849546310018e29afce46401d487d533f","ref":"refs/heads/default","pushedAt":"2024-05-10T17:58:36.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"stellaraccident","name":"Stella Laurenzo","path":"/stellaraccident","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/494208?s=80&v=4"},"commit":{"message":"Add fork roll.","shortMessageHtmlLink":"Add fork roll."}},{"before":"eb33134d8e8721357270a05b54f5f8fec4fb5f5f","after":"49d93822e1301dc8f29d88f1af0c989ed8da4ee3","ref":"refs/heads/main","pushedAt":"2024-05-09T17:01:09.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"nod-automator","name":null,"path":"/nod-automator","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/140433532?s=80&v=4"},"commit":{"message":"Add link to StableHLO page on official OpenXLA website (#2299)","shortMessageHtmlLink":"Add link to StableHLO page on official OpenXLA website (openxla#2299)"}},{"before":"f3835c5c1e1821a7b34450cd643b0301f95ed133","after":"eb33134d8e8721357270a05b54f5f8fec4fb5f5f","ref":"refs/heads/main","pushedAt":"2024-05-09T01:24:31.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"nod-automator","name":null,"path":"/nod-automator","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/140433532?s=80&v=4"},"commit":{"message":"Bump patch version after integrate 0.19.10 -> 0.19.11 (#2307)","shortMessageHtmlLink":"Bump patch version after integrate 0.19.10 -> 0.19.11 (openxla#2307)"}},{"before":"d5368c92c8ccc4b86af40f94db75a28beb7d726d","after":"f3835c5c1e1821a7b34450cd643b0301f95ed133","ref":"refs/heads/main","pushedAt":"2024-05-09T00:03:05.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"nod-automator","name":null,"path":"/nod-automator","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/140433532?s=80&v=4"},"commit":{"message":"dynamic_iota op : match ODS description with the spec (#2305)","shortMessageHtmlLink":"dynamic_iota op : match ODS description with the spec (openxla#2305)"}},{"before":"ab780f6990ed63b9b66df44a0c47eae7c1f69d73","after":"d5368c92c8ccc4b86af40f94db75a28beb7d726d","ref":"refs/heads/main","pushedAt":"2024-05-08T22:01:03.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"nod-automator","name":null,"path":"/nod-automator","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/140433532?s=80&v=4"},"commit":{"message":"Remove dynamic_{iota,reshape} from list of ops to spec (#2303)\n\nThese ops were specced in https://github.com/openxla/stablehlo/pull/2270\r\nand https://github.com/openxla/stablehlo/pull/2284.","shortMessageHtmlLink":"Remove dynamic_{iota,reshape} from list of ops to spec (openxla#2303)"}},{"before":"bac0cd166d8369f8030a285a148409488dd8c89c","after":"ab780f6990ed63b9b66df44a0c47eae7c1f69d73","ref":"refs/heads/main","pushedAt":"2024-05-08T20:01:07.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"nod-automator","name":null,"path":"/nod-automator","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/140433532?s=80&v=4"},"commit":{"message":"Change uses of deprecated method a.cast to cast<>(a) (#2301)\n\nBack porting from the integrate","shortMessageHtmlLink":"Change uses of deprecated method a.cast to cast<>(a) (openxla#2301)"}},{"before":"8ba7728d3fdc3ea882e893ee7e53255c95ee0e5a","after":"bac0cd166d8369f8030a285a148409488dd8c89c","ref":"refs/heads/main","pushedAt":"2024-05-08T17:00:59.000Z","pushType":"push","commitsCount":2,"pusher":{"login":"nod-automator","name":null,"path":"/nod-automator","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/140433532?s=80&v=4"},"commit":{"message":"dynamic_reshape op spec (#2284)\n\n* Constraints in the spec:\r\n```\r\nI1 operand is tensor or quantized tensor\r\nI2 output_shape is 1-dimensional tensor constant of type si64\r\n```\r\n* Test coverage;\r\n```\r\nI1.1 operand is tensor or quantized tensor. (covered by ODS)\r\nI2.1 output_shape is not a 1-dimensional tensor constant of type si64 . (covered by ODS)\r\nC1.1 if !is_per_axis_quantized(operand), element_type(result) = element_type(operand). (Added a verifier logic and a test)\r\nC1.2 if is_per_axis_quantized(operand), element_type(result) = element_type(operand) except\r\n quantization_dimension(operand) and quantization_dimension(result) may differ. (Validated; but test can't be added at this moment because interpreter doesn't support dynamic shapes) \r\nC2.1 size(operand) = size(result). (can't be verified at static time for dynamic ops)\r\nC3.1 quantization constraints. (verifier logic is common for reshape and dynamic_reshape op. Tests are already present for reshape op. Not needed for dynamic_reshape.)\r\nC4.1 rank(result) = size(output_shape) (Already present, validated)\r\n```\r\n\r\n* Reference interpreter:\r\npartial implementation to support only static shapes.\r\n\r\n\r\n\r\n\r\nref:#2267\r\nfixes:https://github.com/openxla/stablehlo/issues/2293","shortMessageHtmlLink":"dynamic_reshape op spec (openxla#2284)"}},{"before":"2eb830a3b2fe8472d8e8b26ed626c9642f64392f","after":"8ba7728d3fdc3ea882e893ee7e53255c95ee0e5a","ref":"refs/heads/main","pushedAt":"2024-05-07T18:01:16.000Z","pushType":"push","commitsCount":2,"pusher":{"login":"nod-automator","name":null,"path":"/nod-automator","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/140433532?s=80&v=4"},"commit":{"message":"Use correct license for LegalizeCompositeToCall (#2297)\n\nThis was added with the wrong license a while ago. I think I copied the\r\nlicense from StablehloAggressiveSimplication.cpp and changed the\r\nauthorship and year, but didn't realize that the actual license\r\ndesignation was wrong.","shortMessageHtmlLink":"Use correct license for LegalizeCompositeToCall (openxla#2297)"}},{"before":"fdfab2e4d0837a2307b9e834494a9535d4d5a996","after":"2eb830a3b2fe8472d8e8b26ed626c9642f64392f","ref":"refs/heads/main","pushedAt":"2024-05-07T16:01:15.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"nod-automator","name":null,"path":"/nod-automator","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/140433532?s=80&v=4"},"commit":{"message":"[RFC] Add batching dims to `stablehlo.gather` and `stablehlo.scatter` specification (#2084)\n\nThis RFC proposes adding batching dimensions (across all operands and\r\nresult) to `stablehlo.gather` and `stablehlo.scatter`, similar to\r\n`lhs_batching_dims` and `rhs_batching_dims` in `stablehlo.dot_general`\r\nPlease provide any feedback you feel is valuable.","shortMessageHtmlLink":"[RFC] Add batching dims to stablehlo.gather and stablehlo.scatter…"}},{"before":"b6e3c984b178537f0b17a77ed32ab27b8e209ce0","after":"fdfab2e4d0837a2307b9e834494a9535d4d5a996","ref":"refs/heads/main","pushedAt":"2024-05-06T23:01:21.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"nod-automator","name":null,"path":"/nod-automator","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/140433532?s=80&v=4"},"commit":{"message":"dynamic_iota op spec (#2270)\n\nConstraints in the spec:\r\n```\r\nI1. output_shape is 1-dimensional tensor constant of type si64\r\nI2. iota_dimension of type si64\r\nC1. 0 <= iota_dimension < size(output_shape)\r\nC2. rank(result) = size(output_shape)\r\n```\r\nTest coverage;\r\n```\r\nI1.1 output_shape is not a 1-dimensional tensor constant of type si64 (covered by ODS) si64 (covered by ODS)\r\nI2.1 iota_dimension is not of type\r\nC1.1 0 <= iota_dimension (Already present)\r\nC1.2 iota_dimension < size(output_shape) (Already present)\r\nC2.1 rank(result) = size(output_shape) (Already present)\r\n```\r\n\r\nReference interpreter:\r\npartial implementation to support only static shapes.\r\n\r\nref: https://github.com/openxla/stablehlo/issues/2267","shortMessageHtmlLink":"dynamic_iota op spec (openxla#2270)"}},{"before":"a0d3bd85315ad9fb4eec4287ec725bd75f78a342","after":"b6e3c984b178537f0b17a77ed32ab27b8e209ce0","ref":"refs/heads/main","pushedAt":"2024-05-06T22:01:19.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"nod-automator","name":null,"path":"/nod-automator","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/140433532?s=80&v=4"},"commit":{"message":"Verify StableHLO Quantized Types (#2203)\n\nfixes https://github.com/openxla/stablehlo/issues/2074\r\n\r\nIt seems there are some use-cases, like quantized reduce wanting to\r\nperform the body computation in a higher bitwidth, which avoids\r\nconstraining and hence is not part of the verification checks.\r\n\r\n```\r\n(C1) num_bits(storage_type) < num_bits(expressed_type).\r\n```\r\n\r\nAdded a ticket to address it separately\r\nhttps://github.com/openxla/stablehlo/issues/2205","shortMessageHtmlLink":"Verify StableHLO Quantized Types (openxla#2203)"}},{"before":"75d2f6c5b3d0d4e0179a2e37e7af58a8213db9ab","after":"a0d3bd85315ad9fb4eec4287ec725bd75f78a342","ref":"refs/heads/main","pushedAt":"2024-05-06T21:02:08.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"nod-automator","name":null,"path":"/nod-automator","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/140433532?s=80&v=4"},"commit":{"message":"Remove unused chlo.dynamic_reshape and downstream ops (#2286)\n\nThe downstream ops are stablehlo.compute_reshape_shape and\r\nstablehlo.cstr_reshapable. These ops were used to implement tf.reshape\r\nbut are no longer needed. (Furthermore, use by a single\r\nframework/backend is usually not sufficient to include an op in\r\nCHLO/SHLO).\r\n\r\nContext: https://github.com/openxla/stablehlo/issues/1991 (P1) -- This\r\nproposal was approved as a part of the Dynamism RFC.","shortMessageHtmlLink":"Remove unused chlo.dynamic_reshape and downstream ops (openxla#2286)"}},{"before":"ab92adeda9119a6c3914cd42367b0a2b70765e91","after":"75d2f6c5b3d0d4e0179a2e37e7af58a8213db9ab","ref":"refs/heads/main","pushedAt":"2024-05-06T18:01:21.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"nod-automator","name":null,"path":"/nod-automator","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/140433532?s=80&v=4"},"commit":{"message":"Remove a few more deprecated method uses (#2285)\n\nI think these were added after/during the first round of changes.\r\n\r\nContext: https://github.com/openxla/stablehlo/pull/2180","shortMessageHtmlLink":"Remove a few more deprecated method uses (openxla#2285)"}},{"before":"40e70d295b6ee6c00a348a008f87f80c83de4f50","after":"ab92adeda9119a6c3914cd42367b0a2b70765e91","ref":"refs/heads/main","pushedAt":"2024-05-03T23:01:15.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"nod-automator","name":null,"path":"/nod-automator","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/140433532?s=80&v=4"},"commit":{"message":"Add helper to verify output shape matches result type (#2269)\n\nAnd use the helper to verify shape operands in DynamicBroadcastInDim,\r\nDynamicIota and DynamicReshape.\r\n\r\nAlso, reorganize the isCompatibleForHloTypeInference logic a little bit\r\nto avoid duplication.\r\n\r\nThis is a follow-up/generalization of\r\nhttps://github.com/openxla/stablehlo/pull/2264.","shortMessageHtmlLink":"Add helper to verify output shape matches result type (openxla#2269)"}},{"before":"e0c6d5d4a7a6caa804d4d7bb6f6a6ca8df704a48","after":"40e70d295b6ee6c00a348a008f87f80c83de4f50","ref":"refs/heads/main","pushedAt":"2024-05-03T20:01:25.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"nod-automator","name":null,"path":"/nod-automator","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/140433532?s=80&v=4"},"commit":{"message":"Update test name to match tested element type (#2282)","shortMessageHtmlLink":"Update test name to match tested element type (openxla#2282)"}},{"before":"c0132de0d56a28df2bf7f65c725b541b8eb62227","after":"e0c6d5d4a7a6caa804d4d7bb6f6a6ca8df704a48","ref":"refs/heads/main","pushedAt":"2024-05-03T17:01:21.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"nod-automator","name":null,"path":"/nod-automator","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/140433532?s=80&v=4"},"commit":{"message":"Use map_to_vector instead of map_range ∘ to_vector (#2281)","shortMessageHtmlLink":"Use map_to_vector instead of map_range ∘ to_vector (openxla#2281)"}},{"before":"41e712268dd697b9d6e07834986bbf9532570c0d","after":"c0132de0d56a28df2bf7f65c725b541b8eb62227","ref":"refs/heads/main","pushedAt":"2024-05-03T00:03:16.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"nod-automator","name":null,"path":"/nod-automator","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/140433532?s=80&v=4"},"commit":{"message":"Use free function instead of method for dyn_cast_or_null (#2280)\n\nThe method variant is deprecated.","shortMessageHtmlLink":"Use free function instead of method for dyn_cast_or_null (openxla#2280)"}},{"before":"e45abc24d0e80714928ed978645809add9ec79c0","after":"41e712268dd697b9d6e07834986bbf9532570c0d","ref":"refs/heads/main","pushedAt":"2024-05-02T22:01:20.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"nod-automator","name":null,"path":"/nod-automator","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/140433532?s=80&v=4"},"commit":{"message":"Ensure result shape is static before folding (#2279)\n\nAdded test case which crashes before this change due to pattern\r\napplication order.\r\n\r\nThere's probably some overly safe checks here (like checking that ops\r\nwhich cannot have dynamic output shapes - broadcast_in_dim /\r\nget_dimension_size) but there's no harm, and its safer if we ever add\r\nany additional constraints on result types that can be folded in the\r\nfuture.","shortMessageHtmlLink":"Ensure result shape is static before folding (openxla#2279)"}},{"before":"de6488a8d88b0c8f954fabd40d0fc75dc1e15765","after":"e45abc24d0e80714928ed978645809add9ec79c0","ref":"refs/heads/main","pushedAt":"2024-05-02T17:01:10.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"nod-automator","name":null,"path":"/nod-automator","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/140433532?s=80&v=4"},"commit":{"message":"Interpreter support for dynamic programs (#2275)\n\nThis PR adds a pass-pipeline to refine arguments and shapes of a program\r\nfollowed by\r\n[StablehloCanonicalizeDynamism](https://github.com/openxla/stablehlo/blob/main/stablehlo/transforms/StablehloCanonicalizeDynamism.cpp)\r\nto replace dynamic StableHLO ops with the corresponding static\r\ncounterparts if applicable* (see note below).\r\n\r\nThe goal is to interpret dynamic programs after applying the above\r\npipeline.\r\n\r\npartially-fixes https://github.com/openxla/stablehlo/issues/2202\r\n\r\n\r\nNote:\r\n\r\nFor a programs like\r\n\r\n```\r\nfunc.func @main(%arg0: tensor, %arg1: tensor<2xi64>) -> tensor {\r\n %0 = stablehlo.dynamic_reshape %arg0, %arg1 : (tensor, tensor<2xi64>) -> tensor\r\n return %0 : tensor\r\n}\r\n```\r\n\r\nwhen applied through the above pass-pipeline with static input shapes\r\nlike `tensor<6xf32>,tensor<2xi64>`, will not be able to replace\r\n`stablehlo.dynamic_reshape` with its static counterpart `reshape` as\r\n[StablehloCanonicalizeDynamism](https://github.com/openxla/stablehlo/blob/main/stablehlo/transforms/StablehloCanonicalizeDynamism.cpp)\r\ncan only do that when the second argument of `stablehlo.dynamic_reshape`\r\nis constant, which is not true in this case.","shortMessageHtmlLink":"Interpreter support for dynamic programs (openxla#2275)"}},{"before":"b725ed9fd312db7e775048630df29e24fa9db15d","after":"de6488a8d88b0c8f954fabd40d0fc75dc1e15765","ref":"refs/heads/main","pushedAt":"2024-05-02T01:23:47.000Z","pushType":"push","commitsCount":2,"pusher":{"login":"nod-automator","name":null,"path":"/nod-automator","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/140433532?s=80&v=4"},"commit":{"message":"Integrate LLVM at llvm/llvm-project@2914a11e3fad (#2266)","shortMessageHtmlLink":"Integrate LLVM at llvm/llvm-project@2914a11e3fad (openxla#2266)"}}],"hasNextPage":true,"hasPreviousPage":false,"activityType":"all","actor":null,"timePeriod":"all","sort":"DESC","perPage":30,"cursor":"djE6ks8AAAAEa7_pUwA","startCursor":null,"endCursor":null}},"title":"Activity · iree-org/stablehlo"}